Improved epicardial cardiac fibroblast generation from iPSCs

Abstract

Since the initial isolation of human embryonic stem cells and subsequent discovery of reprogramming methods for somatic cells, thousands of protocols have been developed to create each of the hundreds of cell types found in-vivo with significant focus on disease-prone systems, e.g., cardiovascular. Robust protocols exist for many of these cell types, except for cardiac fibroblasts (CF). Very recently, several competing methods have been developed to generate these cells through a developmentally conserved epicardial pathway. Such methods generate epicardial cells, but here we report that prolonged exposure to growth factors such as bFGF induces fibroblast spindle-like morphology and similar chromatin architecture to primary CFs. Media conditions for growth and assays are provided, as well as suggestions for seeding densities and timepoints for protein harvest of extracellular matrix. We demonstrate marker expression and matrix competency of resultant cells as shown next to primary human cardiac fibroblasts. These methods provide additional guidance to the original protocol and result in an increasingly stable phenotype.

1. Introduction

Fibrotic diseases are thought to account for 45% of U.S. mortality and include conditions such as heart failure, where the myocardium gradually builds up scar tissue over time and ultimately inhibits mechanical function [1]. Fibroblasts are thought to be at the center of fibrosis, as they secrete the extracellular matrix (ECM) proteins that are crosslinked into a scar. While several molecules and physical signals have been identified as being pro-fibrotic (i.e. stiffness, stretch, TGF-β, IL-4, IL-10, etc), the stimuli that activate these pro-fibrotic gene programs are not fully understood [2–4]; animal models help to understand tissue-level processes such as leukocyte infiltration and gross ECM remodeling, but lack the resolution of a reductionist system. Therefore, in-vitro models have been developed to better understand the cellular crosstalk that occurs between cardiac populations in the context of their microenvironment. While primary cells allow for the use of a human system, they rapidly activate in normal culture conditions [5]. Thus, any human primary fibroblast study is likely to require several patient samples of varying backgrounds, which can confound independent variables being tested, in addition to incurring large costs.

Recently, several groups have worked to overcome these limitations by using pluripotent stem cell-based systems, in which a continuous supply of cardiac fibroblasts can be produced from one donor’s stem cells. Over the last decade, large strides have been made in the production of epicardial cells, which are precursors to the fibroblasts lining the coronary arteries (among other cell types), reducing the amount of growth factors and time required to generate these cells [6–10]. To further differentiate epicardial cells, several groups have proposed using bFGF with different media to accomplish a fibroblast fate [10,11]. A similar bFGF treatment has also been shown to differentiate cardiac progenitor cells directly into fibroblasts by way of second heart field progenitors – a method most likely to accurately model atrial and aortal fibroblasts [12]. However, the left ventricular wall is primarily populated by epicardial fibroblasts that progress through the cardiac progenitor stage, and therefore a Gsk3/Wnt/Gsk3 inhibiting (GiWiGi) protocol is most likely to recapitulate native physiology of ventricular fibrosis in conditions such as heart failure and myocardial infarction [9,13]. Here we focus on the epicardial lineage, particularly as it relates to using cardiac fibroblasts (CFs) for in-vitro disease modeling of the impacts of left ventricular fibrosis.

Two epicardial-derived protocols have recently been published, both yielding cardiac fibroblasts 18 days from initial GSK3 inhibition with CHIR9902 [10,11]. While largely similar though epicardial cell generation, protocols diverge in their approaches to generate CFs with basic fibroblast growth factor (bFGF). A protocol from Bao et al. relies on a serum-free media with low doses of bFGF (10 ng/mL) over 6 days [9], while a protocol from Zhang et al. opts for a proprietary serum-containing medium (i.e., Promocell Cardiac Fibroblast Growth Medium 3) with bFGF and human insulin, which is then supplemented with additional bFGF [10] and TGF-β inhibitor (Fig. 1A). While neither is serum-free, the protocol from Bao et al. is substantially more chemically defined (but uses 20% serum-containing media to neutralize Accutase after passaging) and employs more efficient use of growth factors.

Fig. 1. Comparison of CF differentiation protocols.

A. Schematic of various cardiac fibroblast differentiations, highlighting differences within and between developmental origins. B. Brightfield (left) images of Promocell primary ventricular cardiac fibroblasts and the resultant cells from each protocol after being cultured in Promocell Fibroblast Growth Medium at 25 k/cm² for three days (n = 3, cultured in triplicate, repeated twice). Immunofluorescent staining of CFs (25 k/cm²) after 3 days of culture in assay media for fibronectin-EDA (green), αSMA (red), and DAPI (blue). C. Quantification of cell circularity between fibroblast differentiations. 10 cells were traced in three images for each group, p < 0.05 using a one-way ANOVA. D. Representative western blot for type I collagen, fibronectin-EDA, and beta Actin of mRIPA soluble CF lysate (50 k/cm²) cultured for 3 days in assay media (n = 3, samples differentiated in triplicate). E. Densitometry of western blots, bands normalized to beta actin and primary groups between blots. n = 3, significance is defined as p < 0.05 by one-way ANOVA.

In differentiating CFs from patient-specific iPSCs using the protocols from Bao et al. and Wu et al., we were unable to obtain proper morphologic phenotype (Fig. 1B). However, several modifications to the protocol produced a proper phenotype that yielded a similar chromatin architecture to primary CFs. Specifically, we found that: (1) extending the bFGF treatment of epicardial cells from 6 days to 20 days and (2) adjusting culture and assay mediums improved CF differentiation. While this protocol is longer in duration than the aforementioned methods, it presents a low-serum approach that reliably yields fibroblasts that can be grown for at least 15 passages.

2. Methods and materials

2.1. Ethical compliance and cell lines

The authors have complied with all ethical regulations via a study approved by UCSD (IRB #141315) for all patient-derived iPSCs, which were derived originally by the Scripps Research Institute (IRB #11–5676). Full characterization of these lines is available elsewhere [14]. The authors commercially acquired H9 human embryonic stem cells (ESCs) from WiCell (Madison, WI). Primary cells were obtained from CellBiologics (Cat. H-6049, Chicago, IL) for use in ATAC sequencing and Promocell (Cat. C-12375, Heidelberg, DE) for immunofluorescence and western blot.

2.2. Differentiation components and methods

H9 ESCs and patient-derived iPSCs were maintained in their pluripotent state using mTeSR1 and by passaging with Versene and cell scrapers prior to differentiation. All materials required for differentiation are noted in Table 1. Cells were differentiated into epicardial cells according to the protocol from Bao et al. but with notable exceptions outlined below. In the progenitor stage, changes include the use of 5–6 μM of CHIR99021 on day 0, 2.5 μM of IWP2 on day 3, and 3 μM of CHIR99021 on days 7 and 8 (Fig. 2A). Representative transcription factor expression was confirmed by immunofluorescence and morphology resembled reported images (Fig. 2B). Differentiation of epicardial cells into fibroblasts was achieved by 20 days of 10 ng/mL bFGF treatment in LaSR basal medium (Advanced DMEM/F12, 2.5 mM GlutaMAX and 60 μg/mL ascorbic acid) [15]. Differentiations using Zhang-Kamp and Bao protocols were performed according to their publications [9,12]. To test the Zhang-Wu method, epicardial cells generated from Bao protocol were treated with 10 ng/mL of bFGF (R&D Systems, Cat. 233-FB, Minneapolis, MN) and 0.5 μM A83–01 (Tocris, Cat. 2939, Minneapolis, MN) for six days.

Table 1. Materials checklist.

Materials required for the differentiation process from iPSC to CF, excluding materials required for iPSC maintenance.

Item (Catalogue Number, Manufacturer)	Amount for 1 differentiation (includes 10% extra)	When Needed in the Protocol (Day)
• 6-well low evaporation plate – Matrigel coated (354,277, Corning)	1	Before − 3
• Accutase (1154, Innovative Cell Technologies)	27 mL	−3,6,12
• mTeSR1 Media (85,851, Stem Cell Technologies)	63 mL	−3,−2,−1
• Y27632 (Y-5301, LC Labs)	Depends on Cell Yield	−3, 6, 12
• 15 mL conical tubes (352,095, Falcon)	~20	Several, esp. 3
• 12-well low evaporation plate – Matrigel coated (354,277, 353,043, Corning)	1	−3
• RPMI 1640 Medium (11875–093, Gibco)	93 mL	0,1,3,5
• 36 mM CHIR99021 (SML1046, Sigma Aldrich)	6 μL	0,7,8
• 5 mM IWP2 (S7085, Sellechem)	12 μL	3
• RPMI20 Media (RPMI 1640 + 20% FBS) (11,875, Gibco)	5 mL	6, 12
• LaSR Basal Medium 500 mL Advanced DMEM/F12 medium (12,634,028, Thermo Fisher) 6.5 mL Glutamax (35,050,061, Gibco) 500 μL of 100 mg/mL ascorbic acid solution (36,237, Alfa Aesar)	230 mL	6–32
• 12-well low evaporation plate-0.1% gelatin coated (353,043, Corning, G9391, Sigma)	2	6,12
• bFGF (233-FB, RnD Systems)	2.4 μg	13–32
• CHIR99021 (C-6556, LC Laboratories)	Varies	0,6,7
• Fibroblast Growth Medium 3 (C-23025, Promocell)	2 mL/day	Varies
• CPC Freezing Medium 30% FBS 10% DMSO −5 μM Y27632 (Y-5301, LC Labs)	1 mL/million cells	6
• Epicardial Freezing Medium 30% FBS 10% DMSO −5 μM Y27632 (Y-5301, LC Labs) −500 nM A83 (019001799, Cayman Chemical)	1 mL/million cells	12
• Cryostor CS-10 (07930, Stem Cell Technologies)	1 mL/million cells	12,32
• Mr. Frosty Freezing Container (15–350–50, Fisher Scientific)	1	6,12,32
• A83–01 (9,001,799, Cayman Chemical), 500 μM stock	1μL/mL	Only when expanding Epicardial Cells

Fig. 2. Differentiation and characterization of PSC-derived CFs.

A. Brightfield images (top row, black and white) of cells on days 0, 6, 12, and 32, respectively. Immunofluorescent staining of transcription factors (bottom row, left three), and CF markers (yellow box, colored images) of iPSC-derived lines. Scale bars represent 100 μm. N = 3 performed in parallel in triplicate. B. Immunofluorescent staining of CF markers in assay medium (top row) or growth medium (bottom row), using iPSC-derived CFs (scale bars are 200 μm). C. Media comparisons using Promocell primary CFs (scale bars are 100 μm). N = 3 performed in parallel in triplicate. D. qPCR readout of ACTA2 after 10 ng/mL of TGF-β in 1% serum-containing growth medium. E. Flow cytometry FSC-Area vs αSMA fluorescent intensity of untreated (control) or TGF-β (10 ng/mL) after three days. Antibodies were titrated from 0.625 to 7 μL/100 k cells. Geometric mean (right) of each population and significance calculated using a regression slope test, p < 0.01.

2.3. Cardiac fibroblast phenotyping

2.3.1. Immunofluorescence

Expansion and maintenance of CFs with Fibroblast Growth Medium 3 (Promocell, Heidelberg, DE) and 0.25% Trypsin allowed for expansion through at least 16 passages. CFs of passage 8 or below were cultured in either Fibroblast Growth Medium 3 or RPMI1640 + 10% FBS and 250 μM of ascorbic acid for 3 days and stained for αSMA (ab32575, Abcam, 1:500) as an activation marker, TE-7 (NBP2–50082, Novus Bio, 1:100) and PDGFRα (AF-307-NA, RnD Systems, 1:250), to confirm identity, and fibronectin EDA (NBP1–51723, Novus Bio, 1:200) to confirm matrix competency. WT1 (R&D Systems, Cat. AF5729, Minneapolis, MN) and vimentin (Cell Signaling Technology, Cat. 5741, Danvers, MA) were used as epicardial and post-epithelial to mesenchymal transition markers. TCF21 (PA5–53031, ThermoFisher, 1:100) was used to confirm epicardial lineage and requisite CF transcription factor expression. Samples were blocked in 10% donkey serum, 0.3 M glycine, and 1% bovine serum albumin for 1 h, permeabilized in blocking buffer with 0.1% Triton X-100 for 20 min, stained with primary antibodies for 2 h, and then incubated with secondary antibodies (A21202, A10042, and A21447, Invitrogen) for another two hours. Nuclei were stained with DAPI for 15 min at 1:10,000 dilution in DI water, and three washes were performed between each incubation for 5 min each.

2.3.2. Western blot

To identify proteins by western blot, samples were lysed using mRIPA buffer, collected using cell scrapers, and vortexed every 5 min for 30 min total. Afterward, samples were centrifuged at 23,000 g for 15 min and the supernatant was transferred to a new tube. Protein concentrations were calculated using a bicinchoninic acid assay (23,225, ThermoFisher Scientific), and after denaturing at 95° Celsius for 5 min, 10 μg of protein in 30 μL of RIPA buffer was loaded per lane on a 4–12% BisTris Plus Gel (NW04122BOX, Thermo Fisher) in reducing conditions. Gels were run at 140 V for 55 min and transferred using an iBlot nitrocellulose transfer membrane (IB301001, Thermo Fisher). Membranes were blocked using Azure Blot blocking buffer (AC2190, Azure Biosystems) for 1 h, incubated with primary antibodies (Collagen 1: 14695–1-AP, Proteintech, 1:100, GAPDH: ab8245, Abcam, 1:500, Fibronectin-EDA, NBP1–51723, Novus Bio, 1:2500, Beta Actin, ab-8226, Abcam, 1:500) overnight at 4° Celsius, secondary antibodies (A11374 and A10038, Invitrogen) for 1 h, and imaged using a LI-COR Odyssey (LI-COR, Lincoln, NE).

2.3.3. qPCR

RNA was isolated after washing cells with 1× PBS twice, using Trizol (15596026, ThermoFisher). After 5 min of incubation, cells were scraped, transferred to 1.5 mL Eppendorf tubes, and 0.2 mL of moleculargrade chloroform was added. Tubes were shaken and let equilibrate for 3 min, after which tubes were centrifuged for 30 min at 4 degrees Celsius and 3700 RCF. Aqueous phases were transferred to new tubes and 0.5 mL of isopropanol was added. Tubes were inverted and let sit for 10 min. Tubes were centrifuged again using the same parameters as before, but for 20 min. The supernatant was removed and 1 mL of cold 75% molecular grade ethanol in DEPC water was added and vortexed. Tubes were then centrifuged at 3700 RCF for 12 min, decanted, and allowed to air dry. RNA was then resuspended in DEPC water and purified using the RNeasy Mini Kit (74104, Qiagen). RNA concentration was measured by Nanodrop (ND-2000, ThermoFisher), and 1 μg of RNA was used per reverse transcription reaction using Superscript IV and oligo(dT) (18091050, ThermoFisher). cDNA was stored at 20C prior to amplification. 10uL reactions were performed using 5 ng of cDNA, 1 μМ forward and reverse primers in DEPC water, and 5 μL of Sybr Green (4309155, ThermoFisher). Each primer pair was optimized for melt temperature, and efficiency was validated to be between 80 and 120%. GAPDH primers (ran at 61C) were as follows: F-TCGA-CAGTCAGCCGCATCTTC, R-ACCAAATCCGTTGACTCCGAC, and ACTA2 (ran at 64C) was: F-AGCCAAGCACTGTCAGGAAT and R-CAC-CATCACCCCCTGATGTC. Expression was calculated using 2^-ddCT method using GAPDH as the housekeeping gene.

2.3.4. Flow cytometry

Cells were passaged using Accutase (07922, Stem Cell Technologies), counted using a hemocytometer, resuspended in FACS buffer (1× PBS + 2% BSA w/v), and 100 k cells were added to each well of a 96-well round bottom plate (3799, Corning). Cells were heated at 67 degrees Celsius for 4 min as a dead control and added to the plate. Cells were washed with PBS, centrifuged at 1200 RCF for 1 min, and wrist-flicked to remove supernatant. 100 μL of 1:10,000 Tonbo Ghost Dye Red 780 (13–0865-T100, Tonbo Biosciences) in PBS was added to each well as a viability stain and incubated in the dark on ice for 20 min. Antibody solutions (PDGFRα: AF-307-NA, RnD Systems, cTNT: 130–119–674, Miltenyl Biotech, Nanog: PA5–46891, ThermoFisher, αSMA: IC1420A, RnD Systems) were diluted in FACS buffer according to saturation points determined in previous experiments and kept on ice in the dark. Cells were then washed thrice with 150 μL of FACS buffer, centrifuging, wrist-flicking, and triturating with each rinse. 50 μL of each surface antibody was then added to appropriate wells and incubated in the dark on ice for 60 min. Three more rinses were then performed, and the cells were fixed and permeabilized using Cytofix/Cytoperm (554714, BD Biosciences) in a fume hood in the dark. Following three more rinses, intracellular antibodies and BV421 (705–675–147, Jackson Labs) (the secondary antibody used for PDGFRα), were added and incubated in the dark for 1 h. Compensation beads (01–2222–41, ThermoScientific) for each antibody were added with 30 min of incubation time left. After three more rinses, cells were transferred to FACS tubes and analyzed using an LSRFortessa X-20 Analyzer (BD Biosciences) and FlowJo (BD Biosciences). Forward and side scatter gates were drawn for each sample type and gates were drawn above unstained control samples.

2.3.5. Assay for transposase-accessible chromatin (ATAC) sequencing

ATAC-sequencing was performed on patient-derived iPSC clones to confirm similar chromatin accessibility both between clones and also on primary human cardiac fibroblasts obtained from Cell Biologics. Library preparation and sequencing was performed by the UCSD Center for Epigenomics. ATAC-seq was performed on 50,000 nuclei per sample. Samples were permeabilized in cold permeabilization buffer (0.2% IGEPAL-CA630 (I8896, Sigma), 1 mM DTT (D9779, Sigma), Protease inhibitor (05056489001, Roche), and 5% BSA (A7906, Sigma) in PBS (10010–23, Thermo Fisher Scientific)) for 10 min on a rotator at 4 °C followed by centrifugation for 5 min at 500 g at 4 °C. The pellet was resuspended in cold tagmentation buffer (33 mM Tris-acetate (pH = 7.8) (BP-152, Thermo Fisher Scientific), 66 mM K-acetate (P5708, Sigma), 11 mM Mg-acetate (M2545, Sigma), 16% DMF (DX1730, EMD Millipore) in molecular biology grade water (46000-CM, Corning) followed by incubation with Tagmentation enzyme (FC-121–1030; Illumina) at 37 °C with shaking at 500 rpm for 30 min. Tagmented DNA was purified using MinElute PCR purification kit (28004, QIAGEN). The resulting libraries were amplified using NEBNext High-Fidelity 2× PCR Master Mix (M0541, NEB) with primer extension at 72 °C for 5 min, denaturation at 98 °C for 30 s, followed by 8 cycles of denaturation at 98 °C for 10s, annealing at 63 °C for 30s and extension at 72 °C for 60s. After purification of amplified libraries using MinElute PCR purification kit (28004, QIAGEN), double sided size selection was performed using SPRIselect beads (B23317, Beckman Coulter) with 0.55× beads and 1.5× to sample volume. Libraries were sequenced on a NextSeq500 (Illumina). Adaptor-trimmed fastq files were aligned to hg38 by Bowtie2 [16] using parameters “-X2000–mm–local”. After filtering via samtools [17] with “-q 30 -F 1804 -f 2,” only primary and properly mated reads remained. PCR duplicates were removed by using “markduplicate” from Picard tools (http://broadinstitute.github.io/picard/). The remaining mapped reads were shifted +4 bp and 5 bp for “+” and “-” strand respectively to adjust for Tn5 dimer so that the first base of each reads represents the cutting site. Then peak calling was performed by using MACS2 [18], all through the standard UCSD Epigenetics workflow. Peaks were called using the following settings: callpeak “-f BAMPE -g dm - -q 0.01 –nomodel –shift -100 –extsize 200 –keep-dup all”. The output “narrowPeaks” were further filtered to remove blacklisted regions. The detailed pipeline can be accessed at https://github.com/epigen-UCSD/atac_seq_pipeline. XLS and sorted bam files were submitted to Diffbind [19] in R for differential accessibility testing using an FDR cutoff of 0.01. Differentially accessible regions were submitted to GREAT [20] for transcription start site distance determination and ontology annotation. Data generated in this study was deposited to NCBI under GEO GSE167368. We do not impose any restrictions on data availability.

To generate pseudobulk ATAC-seq data from patient-derived cardiomyocytes and cardiac fibroblasts for comparison with bulk ATAC-seq data from our differentiation protocol, we acquired bed files corresponding to aggregated open chromatin reads from atrial cardiomyocytes, ventricular cardiomyocytes, and cardiac fibroblasts from four cardiac chambers of four human donors (available from http://catlas.org/humanheart under “Bed files”) [21]. We next used cellular barcodes to assign reads to specific donors and heart chambers based on each read’s barcode identifier. These final bed files, corresponding to either cardiomyocytes or cardiac fibroblasts from a specific donor and heart chamber, were used as pseudobulk inputs for downstream comparative analysis. To generate genome browser tracks, we converted pseudobulk bed files to bedgraph format using BEDtools [22] via the “genomecov” command with the “scale” option set to 10⁶ / total number of reads in pseudobulk bed file. Bedgraph files were converted into bigwig format using the “bedGraphToBigWig” tool.

3. Protocol

3.1. Cardiac progenitor differentiation with Gsk3 inhibitor and Wnt inhibitor

• Day – 3

  • 1
    Culture hPSCs on Matrigel-coated 6-well plates in mTESR1 medium to 80–90% confluence (cells within colonies should still be uniform in morphology and colonies should not be touching). Aspirate medium, wash with 1mL of PBS per well, and add 1 mL of room-temperature Accutase to each well. Incubate at 37° Celsius, 5% CO2 for 5 min.
  • 2
    Add 1 mL of mTeSR1 media to each well of the 6-well plate and pool all the cells in a 15 mL conical tube. Mix and count cell number on hemocytometer. Centrifuge the cells at 200 g for 5 min at room temperature. Pipetting 3× per well with a P1000 pipette may improve count accuracy.
  • 3
    Aspirate the supernatant, resuspend the PSCs in mTeSR1 + 5 μM Y27632 (ROCK inhibitor) at a density of 0.25–0.5 million cells/mL, and seed 1mL of cell suspension per well of a 12-well Matrigel-coated plate. This constitutes day –3.

    1. Use 0.25 M/well for PSC lines that divide rapidly (passage every 3 days) and 0.5 M/well for slower cells (passage every 4–5 days)
• Day –2 and –1

  • 4
    On day –2 and –1, aspirate the medium and feed cells with with 2 mL of mTeSR1 per well of the 12-well plate.
• Day 0

  • 5
    On day 0, add 4 μL of 36 mM CHIR99021 (GSK3 inhibitor) to 24 mL of RPMI basal medium per plate being differentiated. Aspirate the old medium and then add 2 mL of the RPMI + CHIR99021 solution to each well. Record the time.

    1. This concentration is varies by PSC line, we have used 5–6 μM.
• Day 1

  • 6
    Exactly 24 h later (day 1), aspirate the medium from each well and replace it with 2 mL of room-temperature RPMI basal medium.
• Day 3

  • 7
    48 h later (day 3), prepare combined medium as follows: For each well, aspirate 1 mL of conditioned media and add 1 mL of freshly prepared RPMI basal medium in a 15 mL conical tube (this is the combined medium). Alternatively, combined medium can be prepared in a plate. Add 1 μL of 5 mM IWP2 (final concentration is 2.5 μM) to each conical tube or well of the combined medium plate. Gently shake the cell-containing plate to suspend cell debris and aspirate the remaining 1 mL of medium per well, and then add the 2 mL of combined medium + IWP2 to each well.
• Day 4

  • 8
    Thaw 50 mL of Fetal Bovine Serum
• Day 5

  • 9
    For day 5, aspirate the medium from each well and add 2 mL of room-temperature RPMI basal medium to each well. Return to incubator. Coat a 12-well plate with 0.1% gelatin (250 uL/well) for D6, prepare RPMI20 and LaSR media (see Table 1).

    1. Note: We do not recommend freezing at CPC, but instead waiting for ProEpicardial cells to freeze. If you are not freezing now, you can usually seed 1–3 plates on D6.

3.2. Directed differentiation of cardiac progenitors into epicardial cells

• Day 6

  *If freezing, prepare freezing media first* On day 6, aspirate the medium, rinse with 1mL of PBS/well, and add 1 mL of Accutase per well in a 12-well plate, and incubate the plate for 5 min. Prep tubes for step 10 during incubation.

  • 10
    Pipette slowly10 times with a P1000 tip to singularize the cells, and then transfer 1 mL of the cell mixture to a 15 mL conical tube. Quench the Accutase with 2x cell volume of RPMI20 medium.
  • 11
    Count the cells with a hemocytometer, centrifuge the cells at 200 g for 5 min at room temperature, and aspirate the supernatant. This is a potential freezing point, see freezing/thawing protocol for additional steps – these are cardiac progenitor cells.
  • 12
    Resuspend the cell pellet in LaSR basal medium +5 μM Y27632 (Rock inhibitor) + at a concentration of 60,000 cells/mL, and seed 1mL of cell suspension per well onto a gelatin-coated 12-well plate Incubate overnight.
• Days 7 and 8

  • 13
    On days 7 and 8, prepare 12mL of LaSR basal medium +3 μM CHIR99021 (this is equivalent to 1 μL of CHIR in 12 mL media) per plate being differentiated. Aspirate the old medium and replace it with 1 mL of the LasR + CHIR solution.
• Day 9–11

  • 14
    On days 9–11, aspirate the medium and replenish it with 1 mL of room-temperature LaSR medium per well. On D11, make RPMI20 and gelatin coated 12-well or 6-well plates. Secure a Mr. Frosty for D12 if freezing.

3.3. Differentiation of epicardial cells into fibroblasts

• Day 12

  *If freezing, prepare Epicardial Freezing media first*

  • 15
    At this point you should have epicardial cells.

    1. On day 12, aspirate the medium, rinse with 1mL of PBS, and add 1 mL of room-temperature Accutase/well of the 12-well plate, and incubate the plate for 15 min.
    2. Pipette 10 times with a P1000 tip to singularize the cells, then transfer the cell mixture to a conical tube containing 2x the cell suspension volume of of RPMI20 medium to quench the Accutase.
    3. Count the cells with a hemocytometer, centrifuge the cells at 200 g for 5 min at room temperature, and aspirate the supernatant. This is a freezing point of epicardial cells.
    4. Resuspend the epicardial cells and seed onto a gelatin-coated cell culture dish at a density of 60,000–80,000 cells/cm² in LaSR basal medium supplemented with 5 μM Y27632 (ROCK Inhibitor). Incubate overnight to allow for cell attachment.
  • 16
    The next day and each day (days 13–32) thereafter, aspirate the medium from each well of the 12-well plate, and add 1 mL per well of room-temperature LaSR basal medium with 10 ng/mL bFGF for fibroblast differentiation (1:1000 dilution). You can also do this with 4 mL of media per well in a 6-well plate, but we suggest a slightly higher seeding density with the larger size wells.
  • 17
    On day 32, trypsinize cells (250 μL or 125 μL of 0.25% trypsin per well of 6-well or 12-well plate, respectively) and plate on tissue culture plastic 6-well plates at 20 k cells/cm² in Promocell Fibroblast Growth Medium 3. Cells should maintain proper phenotype over 8–10 passages but not assemble matrix. For matrix assembly or other assay applications, culture cells with RPMI +10% FBS + 50uM ascorbic acid one day after passaging.

3.4. Freezing protocol

1. After dissociation with either Accutase or Trypsin, resuspend the cardiac progenitor cells (from step 11) or epicardial cells (from step 15) at a density of 1 × 10⁶ cells/mL in Cryostor for maximum viability or CPC/EpiC Freezing Medium.

2. Add 1 mL of the cell suspension in to each cryovial, and freeze them using a Mr. Frosty container at −80 °C for at least 2 hours.

3. Once frozen, transfer the cryovials to liquid nitrogen.

3.5. Thawing protocol

1. Incubate the vial in a 37 °C water bath until a pencil eraser-sized ice pellet remains.

2. Gently pipette the thawed cardiac progenitor cells or proepicardial cells into a 15 mL conical tube containing 9 mL of high glucose DMEM or RPMI1640 with 10% FBS. Count the cells using a hemocytometer while performing step 3.

3. Centrifuge the cells at 200 g for 5 min at room temperature, and then aspirate the supernatant.

4. Gently resuspend the cells in LaSR basal medium with 5 μM Y27632 (ROCK Inhibitor) and plate in a gelatin-coated 12-well plate at a density of 0.5 million cells/cm². For epicardial cell thawing, add 0.5 μM A83–01 to prevent premature differentiation.

5. The next day, areplace the media with 2 mL of room-temperature LaSR basal medium. Expand if necessary, then continue with the main protocol.

4. Results and discussion

While several methods have been previously published to generate cardiac fibroblasts, we were only able to successfully reproduce the Zhang and Kamp method [12], which progresses through second heart field progenitors, and yield a cell that morphologically resembled CFs (Fig. 1A–B). We therefore sought to develop a differentiation protocol that resembles primary phenotype while progressing through the epicardial stage. We found that the Bao [9] and Zhang/Wu [10] protocols yielded cells with circular morphology (Fig. 1B, left) and poor matrix assembly (Fig. 1B, right). Quantitatively, these morphological differences were represented by large differences in circularity – canonical 2D fibroblast morphology is spindle-shaped and the existing epicardial protocols produced a more rounded shape (Fig. 1C). Analysis of soluble protein by western blot demonstrated that all protocols including our iCF method produced statistically similar amounts of Fibronectin-EDA, i.e., fibronectin containing an extra A-type domain prevalent in CFs (Fig. 1D,E). When probing for type I collagen, a key protein for modeling scar remodeling, the iCF protocol presented here had the highest type I collagen expression (Fig. 1D,E), akin to primary cells. To better understand how differences between protocols could result in morphological and matrix production differences, we dissected specific steps in select protocols. We found that the addition of TGF-β inhibitor in the Zhang/Wu [10] protocol inhibited epithelial-to-mesenchymal transition (EMT), i.e., punctate nuclear WT1 staining after 6 days of treatment became diffuse in assay media (RPMI, 10% FBS and 100 mg/L Vitamin C; Fig. S1A). In concert with this, vimentin was almost non-existent immediately after differentiation but emerged after culture in assay media, suggesting that EMT was spurred by serum and lack of TGF-β inhibition. For these reasons, we decided to compile our own modifications to the Bao protocol, better reproducing primary cell phenotype, i.e., the iCF protocol described above.

Our modified differentiation timeline (Fig. 1A) yields cardiac progenitor, epicardial cells, and ultimately generates cardiac fibroblasts that express hallmark proteins (Fig. 2A). While many different CF markers have been reported, each has caveats in either specificity or persistence; for example, CD90 is often used as a fibroblast marker, but only marks a subset of CFs and expression decreases over several passages [23]. Fibroblast-specific protein 1 (FSP-1)–another reported CF marker–is also expressed on endothelial cells [24]. We found that the mesenchymal marker Fibroblast TE-7 reliably labels both primary and iPSC-derived CFs (Fig. 2A), though the specific antigen is unknown.

Another hallmark of this protocol is the subsequent ability to expand cells when grown in Promocell Fibroblast Medium 3. This medium was excellent at maintaining proliferation, morphology, and suppression of CF activation (i.e., αSMA positivity; Fig. 2B), but it surprisingly suppressed fibronectin assembly. When assessing CF phenotype, cells were transitioned to assay medium containing RPMI with 10% fetal bovine serum and 50 μg/mL ascorbic acid. This permitted fibronectin assembly and production of collagen 1, though spurring αSMA activation. Culture expanded CFs, when plated at 2 × 10⁴ cells/cm² and then switched to assay medium for 3 days (with daily media changes), produced robust extracellular matrix assembly that was easily quantifiable by immunofluorescence or western blot (Fig. 2B,C, S1B). In response to TGF-β stimulation, fibroblasts should upregulate αSMA as they begin to adopt a myofibroblast phenotype. We found that 10 ng/mL TGF-β was able to upregulate αSMA in our cells at the mRNA and protein levels (Fig. 2D,E). Similar to the Zhang/Wu and Zhang/Kamp findings, we observed low levels of αSMA protein expression in the absence of TGF-β stimulation, but fluorescent intensities increased drastically with dosing. We observe a strong correlation between the peak mRNA level of ACTA2 (approximately twice the level of untreated) and the geometric mean of protein fluorescence by flow cytometry (also approximately 2-fold). Zhang/Wu report a ~1.5-fold upregulation of ACTA2 following 48 h treatment of 5 ng/mL of TGF-β, and Zhang/Kamp demonstrate a 5% increase in high αSMA+ cells following two-day treatment at 10 ng/mL, suggesting both of these cell products may also be TGF-β responsive.

In-vitro studies have also demonstrated that bona-fide cardiac fibroblasts require the TCF21 transcription factor, as knockout of this protein absolves the heart of any CF populations [25]. Since epicardial populations were derived from WT1+ cells (Fig. 2A), it is unsurprising that all derivatives also express and nuclear localize the epicardial marker TCF21 (Fig. 3A). Zhang-Kamp also report upregulation of the TCF21 mRNA toward the end of the differentiation, and we confirm these findings at the protein level. As the cells differentiate from the epicardial stage, they must undergo EMT to establish mesenchymal fate commitment. Vimentin, an intermediate filament that is indicative of EMT [26], was also stained and found to be highest in Primary cells as well as the Zhang-Kamp [12] and this protocols (Fig. 3B). Reduced vimentin expression in the Bao and Zhang-Wu protocols suggests that they have incompletely undergone the transformation. Furthermore, PDGFRα, a key protein required for CF development and survival [27], was found to be highly expressed by primary, Zhang-Kamp, and Whitehead CFs, and to a lesser extent in Bao and Zhang-Wu cells (Fig. 3A–C), depending on the protein quantification technique. However when comparing these cells to other cardiac lineages, e.g., cardiomyocytes, and to their parental line, i.e., iPSCs, no cells were found to significantly express cTnT and Nanog, respectively, since none of the CFs demonstrated a significant right-shift in fluorescence histograms from flow cytometry (Fig. S1C).

Fig. 3. Vimentin and PDGFR α expression are hallmarks of differentiated CFs.

A. Immunofluorescent staining of primary and derived CFs. DAPI, PDGFRα, and TCF21 were co-stained (left) and vimentin is shown with DAPI background. B. Quantification of vimentin and PDGFRα intensity per cell from immunostained samples. N = 3 samples, 3–4 images per sample. Testing performed using a non-parametric one-way ANOVA, p < 0.05. Tukey post-hoc test performed between primary (red lines) and Whitehead samples (blue lines) vs others. C. Flow cytometry histograms of PDGFRα fluorescence across groups demonstrating gate placement right of unstained controls. Quantification of percent PDGFRα+ cells (right) over samples.

To further validate protocol efficacy and because chromatin remodeling is an a priori process to the activation of transcriptional programs, we performed ATAC-sequencing on iPSC lines from two patient clones through each differentiation stage and found that iPSC-CFs clustered with the primary cell samples from Cell Biologics (Fig. 4A). This suggests that chromatin architectural remodeling through the differentiation closely resembles that of native CFs. As expected, each stage of the differentiation also clustered together, showing a continuous trajectory in two principal components.

Fig. 4. ATAC sequencing of differentiation stages.

A. Principal component analysis and correlation heatmap of iPSC-derived cells and Cell Biologics Primary CFs. Primary cells are from the same donor cultured for either 1 or 4 passages. N = 2 per condition, once per iPSC clone for differentiated cells. B. Number of differentially accessible regions from HOMER DESEQ2 output (q < 0.05) when comparing sample groups (n = 2 for differentiated cells and Cell Biologics samples, and N = 3 for primary donor samples from Hocker et al.). There was a total of 70,507 peaks in the merged peakset. C. ATAC sequencing principal component analysis including pseudo-bulk samples from Hocker et al. Cell types clustered together and labels were color coded based on cell identity while point color designates origin of cell, either primary or iPSC-derived. D. Integrative Genomics Viewer (IGV) snapshots of αSMA gene (ACTA2) peaks, including iPSC-derived (n = 2 per cell type), Cell Biologics (n = 2), and ventricular primary pseudo-bulk cardiomyocyte and fibroblast cells (n = 3) from Hocker et al. Tracks displayed are reads per genomic content (RPGC) normalized bigwig files and color-coded by cell type.

Since primary cells obtained from commercial vendors expand primary cells for several passages, we compared our iCFs with primary human cardiomyocyte and cardiac fibroblast pseudobulk single nucleus ATAC sequencing profiles from Hocker et al. [21]. We found that iCFs were most similar to Cell Biologics ventricular cardiac fibroblasts, though all fibroblast groups clustered together and were distinct from cardiomyocyte and stem/progenitor cell populations (Fig. 4B–C). Lastly, since αSMA accessibility precedes the ability of a fibroblast to become activated, we compared accessibility to this genomic region across all ventricular and iPSC-derived samples and found similar accessibility between primary, cell-line, and iCF fibroblasts (Fig. 4D).

In summary, this protocol presents an improved approach to generate CFs for modeling left ventricular fibrosis. We present methods to preserve developmental lineage accuracy via epicardial fate, deliver physiological dosing of bFGF in reduced serum conditions, and greatly improve phenotype of resulting cells. An accurate model of matrix production is required to recapitulate native physiology in a dish, particularly cell adhesion, migration, and substrate stiffness; these cells can be used in downstream applications such as co-culture, 3D matrix assembly, and drug discovery while evading the limitations of primary cells (i.e., low Hayflick limit and varying genetic backgrounds) and murine models.

This breakthrough is particularly important as the field begins to incorporate more complex in-vitro models of disease and multicellular communication before testing in animal systems. Traditionally, mice or rats have been used to model cardiac disease, but they have several limitations: mice hearts beat at approximately 10 times the rate of humans and do not always generate human-like pharmacological responses [28,29]. Rats also have a significantly higher heart rate (approx. 330–480 beats per minute), and while generally better recapitulating human pathological processes, lack many of the genetic tools and strains that make mice an attractive model [29,30]. Though larger mammalian organisms better bridge the gap between animal and human physiology, they are often cost-prohibitive and have even fewer molecular tools than their murine counterparts. Finally, with the emerging understanding of the vast regulatory roles noncoding RNAs play in many diseases and homeostatic processes, model organism genomes often do not contain these sequences and knock-in models fail to capture clinical phenotypes [31–34]. Therefore, in-vitro modeling presents the opportunity narrow therapeutic targets and regenerative approaches prior to more rigorous testing in animals - though in-vitro systems are only beginning to emerge as meaningful screens with translational promise.

Perhaps one of the largest limitations of fibrotic disease modeling-in-a-dish is the inability of fibroblasts to assemble collagen fibrils in 2D. For this reason, many 2D systems measure collagen production by western blot, mRNA, or ELISA of soluble collagen – staining for collagen only resolves intracellular protein and does not represent the bona fide scar manufacturing capabilities of fibroblasts. To circumvent this limitation, groups have investigated using macromolecular crowding, which involves the use of high molecular weight polymers to stabilize BMP1 to cleave procollagen domains and allow for assembly [35]. Thus, fibrotic screens can be performed in 96-well plate formats and be read using a plate reader to quickly screen compounds. Yet even for this method, a major limitation is that different polymers yield different collagen organization and responses to pro-fibrotic agonists [36]. Yet another alternative, though much lower throughput, is to develop multicellular cultures or organoids. This method enables both fibril formation as well as crosstalk between multiple cell types, which may synergistically respond to physical or biological stimuli. In the context of iPSC differentiations, 3D multicellular platforms have also been shown to enhance calcium handling and maturity of cardiomyocytes, as well as their ability to recapitulate congenital heart disease [37–39]. Thus, albeit more time intensive, these in-vitro approaches are more likely to provide anti-fibrotic insights that translate to human therapies than traditional 2D, macromolecular crowding, or animal models. We hope that this differentiation protocol will enable such studies through the development of patient-specific cardiac tissues (now including fibroblasts) to interrogate human pathologies.